Prenatal exposure to air pollution as a potential risk factor for autism and ADHD

Environment International 133 (2019) 105149

Contents lists available at ScienceDirect

Environment International journal homepage: www.elsevier.com/locate/envint

Prenatal exposure to air pollution as a potential risk factor for autism and ADHD

T

Anna Oudina,b, Kasper Frondeliusa, Nils Haglundc, Karin Källénd, Bertil Forsbergb, ⁎ Peik Gustafssonc, Ebba Malmqvista, a

Division of Occupational and Environmental Medicine, Department of Laboratory Medicine, Lund University, Sweden Section of Sustainable Health, Department of Public Health and Clinical Medicine, Umeå University, Sweden c Child and Adolescent Psychiatry, Department of Clinical Sciences Lund, Lund University, Sweden d Centre of Reproduction Epidemiology, Tornblad Institute, Department of Clinical Sciences, Lund University, Sweden b

A R T I C LE I N FO

A B S T R A C T

Handling Editor: Hanna Boogaard

Genetic and environmental factors both contribute to the development of Autism Spectrum Disorder (ASD) and Attention-Deficit/Hyperactivity Disorder (ADHD). One suggested environmental risk factor for ASD and ADHD is air pollution, but knowledge of its effects, especially in low-exposure areas, are limited. Here, we investigate risks for ASD and ADHD associated with prenatal exposure to air pollution in an area with air pollution levels generally well below World Health Organization (WHO) air quality guidelines. We used an epidemiological database (MAPSS) consisting of virtually all (99%) children born between 1999 and 2009 (48,571 births) in the study area, in southern Sweden. MAPSS consists of data on modelled nitrogen oxide (NOx) levels derived from a Gaussian dispersion model; maternal residency during pregnancy; perinatal factors collected from a regional birth registry; and socio-economic factors extracted from Statistics Sweden. All ASD and ADHD diagnoses in our data were undertaken at the Malmö and Lund Departments of Child and Adolescent Psychiatry, using standardized diagnostic instruments. We used logistic regression analyses to obtain estimates of the risk of developing ASD and ADHD associated with different air pollution levels, with adjustments for potential perinatal and socio-economic confounders. In this longitudinal cohort study, we found associations between air pollution exposure during the prenatal period and and the risk of developing ASD. For example, an adjusted Odds Ratio (OR) of 1.40 and its 95% Confidence Interval (CI) (95% CI: 1.02–1.93) were found when comparing the fourth with the first quartile of NOx exposure. We did not find similar associations on the risk of developing ADHD. This study contributes to the growing evidence of a link between prenatal exposure to air pollution and autism spectrum disorders, suggesting that prenatal exposure even below current WHO air quality guidelines may increase the risk of autism spectrum disorders.

1. Introduction Autism Spectrum Disorder (ASD) and Attention-Deficit/ Hyperactivity Disorder (ADHD) are neurodevelopmental disorders associated with broad functional impairments that can substantially affect individuals' and families' quality of life (Klassen et al., 2004; Lee et al., 2008; van Heijst and Geurts, 2015). The etiology behind these conditions is largely unknown, although both environmental and genetic factors play a role (Thapar et al., 2013; Lai et al., 2014; Sandin et al., 2017). Environmental determinants alone are suggested to account for 10–40% of the risk for ADHD (Sciberras et al., 2017) and around 40% for autism (Hertz-Picciotto et al., 2006). ⁎

Air pollution is recognized by the World Health Organization (WHO) as one of the biggest health threats of our time (Cohen et al., 2017). This particular environmental hazard might also have an effect on the central nervous system, (Costa et al., 2017), making it a potential risk factor for developing ASD and ADHD. Animal models have indicated that prenatal exposure to high levels of air pollution causes neurotoxicity (Allen et al., 2017; Costa et al., 2017). Children highly exposed to air pollution have also shown brain imbalances and a breakdown of the blood-brain barrier (Calderon-Garciduenas et al., 2015a; Calderon-Garciduenas et al., 2015b). The known biological effects of air pollution are also in line with the multiple hypothetic mechanisms suggested for ASD and ADHD development (Bolton et al.,

Corresponding author at: Occupational and Environmental Medicine, Lund University, Medicon Village, Scheelevägen 2, Building 402, 221 85 Lund, Sweden. E-mail address: [email protected] (E. Malmqvist).

https://doi.org/10.1016/j.envint.2019.105149 Received 17 January 2019; Received in revised form 20 August 2019; Accepted 2 September 2019 0160-4120/ © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).

Environment International 133 (2019) 105149

A. Oudin, et al.

2012; Allen et al., 2017). Moreover, recent epidemiological studies have found associations between exposure to air pollution and an elevated risk of developing ASD in the United States of America (USA) (Volk et al., 2013; Kalkbrenner et al., 2015; Raz et al., 2015; Talbott et al., 2015; Weisskopf et al., 2015; Flores-Pajot et al., 2016; Yang et al., 2017; Goodrich et al., 2018; Kalkbrenner et al., 2018; Kerin et al., 2018), Taiwan (Jung et al., 2013), and Israel (Raz et al., 2018). These findings have not been reproduced in most European cohort studies investigating prenatal air pollution and autistic traits (Guxens et al., 2016) or ASD (Gong et al., 2014; Gong et al., 2017). However, an association between air pollution and clinical diagnoses of ASD was recently confirmed by a Danish study (Ritz et al., 2018). Fewer articles exist on ADHD and air pollution; most of which relied on teacher- or parent-reported symptoms of ADHD. This body of literature presented contradicting results, as some authors did not find an association (Gong et al., 2014; Forns et al., 2018; Myhre et al., 2018), and others did (Perera et al., 2012; Yorifuji et al., 2017). Furthermore, studies on air pollution and attention functions in Spain and the US have found prenatal exposure to air pollution to be associated with poorer scores in attentiveness (Chiu et al., 2016; Sentis et al., 2017). The one study using clinical diagnosis found an association between air pollution and ADHD development (Min and Min, 2017). There is need to elucidate whether ASD and air pollution findings are replicable in a European setting. More studies are also needed regarding the relationship between air pollution and ADHD. Indeed, those minimizing exposure misclassification, considering the impact of comorbidities, and using population-based samples and clinical diagnoses have been urged (Yang et al., 2017). Other recommendations by the scientific community include increasing power with larger study sizes, tracking residential mobility and incorporating methods for capturing local and regional exposure levels (Flores-Pajot et al., 2016). In this study, we had the possibility to perform population-based research with high quality data on ASD and ADHD, air pollution concentrations with high spatial resolution, and potential confounders, such as other neurodevelopmental disorders, perinatal factors, and socioeconomic status (SES) from nation-wide registers.

Fig. 1. Map of the county of Scania in southern Sweden with modelled levels of NOx for the year 2009 and locations of Malmö and Lund Departments of Child and Adolescent Psychiatry (study clinics).

We identified ASD cases as children with an F84 diagnosis code, which represents all Pervasive Developmental disorders (including i.e. Asperger's syndrome). This is characterized by one or more of the following areas of psychopathology: qualitative abnormalities in reciprocal social interactions, in patterns of communication, and as a restricted, stereotyped and repetitive repertoire of interests and activities. All of which affect the individual's functioning in everyday situations. These symptoms do not have to be present by a specified age, but children usually receive a diagnosis around 7 years of age (Haglund and Kallen, 2011). Furthermore, we restricted cases using the F84.0 code for Childhood autism (excluding i.e. Asperger's syndrome). The criteria for F84.0 is abnormal functioning in all three areas of psychopathology with early onset. Specifically, symptoms need to be present before 3 years of age. To ensure that the core symptoms of autistic disorders are present, the team of professionals use both Autism Diagnostic Observation Schedule–Generic (ADOS-G) (Lord et al., 2000) and the Autism Diagnostic Interview–Revised (ADI-R) (Lord et al., 1994) for the majority (75%) of cases. Additionally, children with ASD can be identified through Child Habilitation Centers, which support all children and adolescents below 18 years of age who fulfil the criteria for ASD and live in the area. In reviewing data from these centers, we found three additional children not previously diagnosed at the Malmö and Lund Departments of Child and Adolescent Psychiatry. A child with suspected attention difficulties, hyperactivity and/or difficulties with impulse control is generally referred to the Departments of Child and Adolescent Psychiatry by a special education teacher and a school psychologist or by their parents. Each child diagnosed with ADHD was assessed by one of the ten experienced clinicians at the Departments of Child and Adolescent Psychiatry in Malmö and Lund using the Diagnostic and Statistical Manual of Mental Disorders (DSM). An ICD10 diagnosis corresponding well with the DSM diagnosis was then used. For ADHD, we included both the broader definition, F90, and the Swedish-specific subtype, F90.0B, requiring hyperactivity/impulsivity with or without clinically significant attention-deficit. We excluded 72 children with a diagnosis in both the F84 and F90 ICD10 groups to reduce potential misclassifications as suggested by Craig et al. (2015).

2. Method 2.1. Study population MAPSS (Maternal Air Pollution in Southern Sweden) is an epidemiological database with > 48,000 births in the Malmö-LundTrelleborg region located within Skåne [Scania], the Southern-most county in Sweden (see Fig. 1). The study population consists of virtually all (99%) children born in Skåne between 1999 and 2009 and has previously been used to study fetal growth and air pollution (Malmqvist et al., 2017). Data concerning diagnoses was checked starting in 1999; however, very few cases prior to 2004 were found, thus children were followed between 2004 and 2016. We used data from Statistics Sweden to assess if the MAPSS children were still living in the catchment areas of the Malmö and Lund Departments of Child and Adolescent Psychiatry during the study period. 2.2. Outcome In this study, we extracted outcome data from the Skåne Healthcare database (SHR). Because health care systems in Sweden are tax-subsidized, readily-available and used by practically all residents, utilizing these healthcare databases does not raise the risk of selection bias, as is the case in many other countries. When a child is suspected of having ASD or ADHD, they are referred to the Departments of Child and Adolescent Psychiatry and examined by a multidisciplinary team. These professionals diagnose cases according to the International Classification of Mental and Behavioral Disorders version 10 (ICD10) and add them into the SHR. 2

Environment International 133 (2019) 105149

A. Oudin, et al.

(Haglund and Kallen, 2011). Moreover, maternal smoking was not seen to be a risk factor for the development of autism (Rosen et al., 2015) but was for ADHD (Thapar et al., 2013). Recognizing these risk factors, we adjusted for sex of the child, maternal age at birth (continuous variable), maternal parity (1, 2, 3 and ≥ 4), maternal smoking during pregnancy (none, 1–9 cigarettes/day, ≥10 cigarettes/day and missing information) and maternal BMI (< 18, 18- < 25, 25- < 30 and ≥ 30 kg/m2) in our main analysis. In the sensitivity analyses, we adjusted for birth weight, gestational age and preeclampsia.

2.3. Air pollution exposure The modelled concentration of nitrogen oxides (NOx) was used as an indicator of combustion-related air pollution from local sources, mainly traffic. NOx levels were modelled using a Gaussian dispersion model, AERMOD, and an extensive emission database on road networks (including type of vehicle, speed, and number of vehicles per minute), small-scale heating, incinerators, industry, non-road machinery, and shipping. We applied a spatial resolution of 500*500 m for the years 1999–2005 and 100*100 m for 2006–2009. Modelled levels have been compared to measured concentrations with good correlations for NOx (Stroh et al., 2007; Stroh et al., 2012). Further, this air pollution model has been used extensively in previous studies (Stroh et al., 2005; Oudin et al., 2009; Malmqvist et al., 2013; Malmqvist et al., 2015; Malmqvist et al., 2017; Frondelius et al., 2018). Modelled NOx concentrations were then linked to the geocoded residential address of each individual mother to obtain the study population's air pollution exposure. We calculated trimester-specific exposure quartiles as well as average exposure levels during the full pregnancy. In the main analysis, we used year-specific quartiles. In an additional analysis, we instead used quartiles for the whole time period, i.e. non year-specific. In main analyses we also studied a continuous measure of NOx (per 10 μg/m3 increase).

2.4.4. Psychiatric unit factors Distance to the psychiatric unit may influence the probability of being diagnosed, and diagnosis may depend on which psychiatric unit is visited. In our particular setting, the two psychiatric units (Lund and Malmö) have clear catchment areas with a travel radius of ≤50 km for any resident. However, it was quite common for patients from other municipalities to seek care at the Lund and Malmö units. We, therefore, had many patients in our data who did not live in the catchment area. In our main analysis, we included patients (and non-patients) from these municipalities too, given that their mothers had lived in the catchment area during pregnancy. In a sensitivity analysis, we limited our investigation to study subjects living in the catchment area. An additional sensitivity analysis was performed to adjust for psychiatric unit (Lund or Malmö). By incorporating birth year into the model as a categorical variable, we could account for time to diagnosis and diagnosis trends in a final sensitivity analysis.

2.4. Potential confounding factors 2.4.1. Parental birth country ASD is over-represented among immigrants from sub-Saharan Africa (Haglund and Kallen, 2011), and health-seeking behavior might differ between Swedes and immigrants (Ivert et al., 2013). Additionally, immigrant and non-immigrant populations are unequally affected by air pollution, with concentrations unevenly distributed in some areas (Stroh et al., 2005). This is often related to socioeconomic status, as discussed below. Specifically, maternal country was aggregated into the following groups: Sweden, Nordic countries except Sweden, European Union member nations (EU28) except the Nordic countries, Europe except EU28 and Nordic countries, the Americas, Africa and Asia.

2.5. Statistical analyses We used logistic regression with ASD and ADHD as outcome variables (four different variables, as defined above) with average residential NOx exposure during the three trimesters and full pregnancy either entered as a continuous or as a categorical variable (quartiles). We ran these analyses in two different sets of models: one crude model and one “adjusted” model. In the adjusted models the following variables were entered: gender of child, maternal age, parity, maternal smoking, maternal BMI, maternal education, disposable income, maternal country of birth. In a cohort setting the odds ratios produced by logistic regression should approximate the Hazard Ratios from a Cox Proportional Hazard Regression. As a sensitivity analysis, we therefore ran the main analyses with Cox Proportional Hazard Regression. Censoring occurred at diagnosis, relocation outside the study area or death, whichever came first. In sensitivity analyses, we adjusted for birth weight, gestational age and preeclampsia. Dose-response curves of the risk estimates versus the NOx concentrations were created using a natural cubic spline for levels between 0 and 30 μg/m3. Above 30 μg/m3 there are very few observations, and data are very skewed. All analysis was done using SAS version 9.4.

2.4.2. Socioeconomic status As in all studies on the long-term effects of air pollution, it is important to distinguish between the effects of socio-economic status (SES) and those from air pollution exposure. We have earlier observed complex patterns between SES and air pollution in this study area (Stroh et al., 2005) suggesting that associations between air pollution and health could be confounded by SES. In the current study, individual-level data on education level as well as household income from Statistics Sweden were used as markers for SES. Maternal education was represented by six categories, although the two highest categories were merged in some of the statistical models, and household disposable income was divided into quartiles.

3. Results

2.4.3. Potential perinatal and maternal risk factors Data on maternal smoking, maternal age and sex of the child were obtained from a local birth registry, Perinatal Revision South (PRS) with nearly full coverage of the population (> 99%). A previous study in this area identified the perinatal and maternal risk factors for developing ADHD and ASD as being maternal age (> 40 years), primiparae (only borderline significant) and male sex (Haglund and Kallen, 2011). Indeed, boys are more often diagnosed with ADHD (Arnold, 1996) and autism (Werling and Geschwind, 2013). Preeclampsia is also a suggested risk factor for both ASD and ADHD (Zhu et al., 2016; Dachew et al., 2018). Some risk factors, however, differed between Asperger's syndrome and Childhood autism cases. For instance, being born preterm (before 37 gestational weeks) and having low birth weight (< 2500 g) was associated with Childhood autism (Wang et al., 2017) and ADHD (Thapar et al., 2013) but not with Asperger syndrome

The mean NOx exposure in the population was 17.7 μg/m3, with a standard deviation of 10.3 μg/m3, ranging from 3.5 μg/m3 to 48.6 μg/ m3. The (non year-specific) quartile limits were 11.3, 16.8 and 23.1 μg/ m3. Age at the time of diagnosis varied between 1 and 17 years. Children with an ASD diagnosis were slightly younger at diagnosis (range = 1–17 years, mean age = 7 years) than children with an ADHD diagnosis (range = 5–17 years, mean age = 10 years). The distribution of the participants' other characteristics was generally quite similar between the children with both ASD and ADHD compared to the whole cohort (Table 1). Although, the proportion of boys was much higher for the children with either diagnosis, especially for ASD. Mothers of children with ADHD seemed somewhat younger than the rest of the cohort, whereas the opposite was seen for mothers with ASD-diagnosed 3

Environment International 133 (2019) 105149

A. Oudin, et al.

Table 1 Participant characteristics, frequencies and column percentages of covariates for all and for the specific outcomes: Childhood autism (F84.0), Autism Spectrum Disorder, ASD (F84), Attention-Deficit/Hyperactivity Disorder, ADHD with hyperactivity/impulsivity (F90.0B) and Attention-Deficit/Hyperactivity Disorder, ADHD (F90). Characteristics

Gender Maternal age (years)

Maternal parity

Mat. Smoking Cigarettes/day

Maternal BMI kg/m2

a

Girl Boy < 20 20-30 30-38 > 38 1 2 3 ≥4a Missing No Cig/Day 1-9 Cig/Day ≥10 Cig/Day < 18.5 18.5- < 25 25- < 30 ≥30 Missing

Childhood autism F84.0 (n=333)

ASD F84 (n=435)

ADHD subtype (F90.0B) (n=516)

ADHD (F90) (n=718)

All (n=48571)

77 (23) 256 (77) 9 (3) 143 (43) 145 (44) 36 (11) 178 (53) 91 (27) 43 (13) 21 (6) 25 (8) 271 (81) 24 (7) 13 (4) 7 (2) 149 (45) 88 (26) 43 (13) 46 (14)

101 (23) 334 (77) 10 (2) 188 (43) 199 (46) 38 (9) 238 (55) 116 (27) 58 (13) 23 (5) 26 (6) 357 (82) 35 (8) 17 (4) 15 (3) 202 (46) 108 (25) 61 (14) 49 (11)

141 (27) 375 (73) 23 (5) 297 (58) 168 (33) 28 (5) 247 (48) 179 (35) 54 (10) 36 (7) 41 (8) 332 (64) 86 (17) 57 (11) 14 (3) 231 (45) 120 (23) 74 (14) 77 (15)

191 (27) 73 (73) 32 (5) 402 (56) 239 (33) 45 (6) 337 (47) 239 (33) 88 (12) 54 (8) 49 (7) 465 (65) 120 (17) 84 (12) 16 (2) 333 (46) 163 (23) 101 (14) 105 (15)

23570 (49) 25001 (51) 850 (2) 21660 (45) 22530 (46) 3531 (7) 22638 (47) 16560 (34) 6084 (13) 3289 (7) 3458 (7) 40521 (83) 3228 (7) 1364 (3) 1145 (2) 26650 (55) 10423 (21) 4435 (9) 5918 (12)

The distribution of this group was quite skewed, the 25th and 50th percentile was 4, the 75th percentile was 5, and the maximum value was 14.

children. Furthermore, there was a larger proportion of first-born children with an ASD diagnosis than was seen in the cohort as a whole. Maternal education and income were markedly lower in mothers with an ADHD-diagnosed child. Additionally, having a mother born in Sweden was more common for children in the ADHD group compared to the whole cohort. Diagnoses seemed to depend on year of birth in the ADHD group, which is not surprising given that diagnoses in this population are highly age-dependent. For those diagnosed with ADHD, all exposure variables seemed to be distributed similarly with respect to the rest of the cohort. Conversely, exposure was generally higher among the ASD group than the whole cohort. Mean levels of NOx varied between the years with mostly a declining trend: average concentrations in 1999 were 21 μg/m3 and 14 μg/m3 in 2009. The results from the logistic regression analysis suggest an association between exposure to NOx during fetal life and ASD diagnosis in our data. For example, an Odds Ratio (OR) of 1.40 was found when comparing first to fourth quartile of NOx exposure in trimester 1 (95% CI: 1.02–1.93) (Table 2). Results were very similar between the trimesters. The OR for a 10 μg/m3 increase in NOx was 1.15 (1.01–1.31) for trimester 1. For trimester 2 and 3, the point estimates were lower (1.11 and 1.10), and not statistically significant. When looking at the dose-response curve (Fig. 2), there seems to be a linear increase in ASD related to air pollution exposure until reaching a NOx concentration of about 25 μg/m3, but with higher concentrations, this trend instead seems to slightly decrease. The analysis including only the Childhood autism subgroup showed similar results to the whole ASD group, but with less statistical precision (Supplementary material, Table 1). For ADHD, however, there did not seem to be an association with prenatal NOx exposure. In the crude analyses, an association in the opposite direction of our hypothesis was shown, but ORs in the adjusted models shifted closer to 1 (Table 3 and Supplementary material, Table 2). Results were very similar between the trimesters. Further, using Cox proportional regression analysis produced similar estimates as logistic regression (data not shown). When including only study subjects residing in the catchment areas, the cohort size was reduced by approximately 8%. Even though the smaller sample size decreased precision, the point estimates were similar to those of the main analysis. Furthermore, adjusting for clinic (Lund or Malmö) only had a marginal effect on the estimates. Further adjustments for preeclampsia, birth weight and gestational age did not seem to influence the findings. Finally, results from the subgroup

Table 2 Odds ratios (ORs) with their 95% Confidence Intervals (CIs) for Autism Spectrum Disorder (ASD F84) in association with NOx exposure during fetal life in crude and adjusted models. Exposurea

Crude modelb N = 47, 865 OR (95% CI)

Adjusted modelc N = 38, 280 OR (95% CI)

Trimester 1 Quartile 1a Quartile 2a Quartile 3a Quartile 4a Linear per 10 μg/m3

1 1.01 1.21 1.47 1.19

(0.75–1.36) (0.92–1.61) (1.12–1.92) (1.06–1.34)

1 1.15 1.34 1.40 1.15

(0.84–1.59) (0.98–1.84) (1.02–1.92) (1.01–1.31)

Trimester 2 Quartile 1a Quartile 2a Quartile 3a Quartile 4a Linear per 10 μg/m3

1 1.04 1.23 1.46 1.16

(0.77–1.39) (0.93–1.62) (1.11–1.91) (1.04–1.30)

1 1.06 1.26 1.35 1.11

(0.77–1.46) (0.92–1.72) (0.98–1.84) (0.97–1.26)

Trimester 3 Quartile 1a Quartile 2a Quartile 3a Quartile 4a Linear per 10 μg/m3

1 1.27 (0.96–1.70) 1.43 (1.08–1.89) 1.48(1.12–1.95) 1.16 (1.03–1.30)

1 1.28 1.41 1.39 1.10

(0.94–1.75) (1.03–1.93) (1.01–1.9) (0.96–1.26)

All pregnancyc Quartile 1a Quartile 2a Quartile 3a Quartile 4a Linear per 10 μg/m3

1 1.22 1.29 1.55 1.20

1 1.29 1.34 1.40 1.14

(0.95–1.76) (0.98–1.84) (1.02–1.93) (0.99–1.31)

(0.92–1.63) (0.97–1.71) (1.18–2.04) (1.06–1.36)

Year-specific quartiles (birth year) with first quartile as reference. The linear estimates are adjusted for birth year (since the quartiles are birth year-specific). c Variables included into the adjusted models: gender of child, maternal age, parity, maternal smoking, maternal BMI, maternal education, disposable income, maternal country of birth. a b

analysis on children of Swedish-born mothers were very similar to the main results. Using year-specific quartiles or not did not influence the results, and the findings were nearly identical when adjusting for birth year.

4

Environment International 133 (2019) 105149

A. Oudin, et al.

(Muhle et al., 2004). However, the underlying causal mechanism is still unknown. More recently, environmental factors have been examined as a potential explanation of ASD's surge among children, particularly those with milder forms of ASD and developmental delay. Discussions surrounding epigenetics have given rise to a hypothesis about genetically predisposed children: an unfavorable environment can exacerbate preexisting vulnerability to ASD and, consequently, manifest into diagnosis (Posar and Visconti, 2017). Another issue to consider is that the criteria for ASD diagnosis has broadened. With this, the classification of children has changed and, ultimately, increased prevalence, which challenges the occurrence of a genuine increase of autistic symptoms among some ASD-diagnosed children (Arvidsson et al., 2018). Considering these complex aspects and the connection between genetics and environment is crucial when interpreting this study's results. Time trends were evident in both exposure and diagnosis, which could in theory be a possible explanation for our findings. We therefore used year-specific quartiles in the main analysis, but the results were nearly identical as when using quartiles based on the whole time period, when also adjusting for birth year. We cannot entirely rule out that time trends might influence the results somehow, but it seems less likely given these outcomes.

Fig. 2. Dose-response curve of the Hazard Ratios (from Cox Regression) versus the NOx concentration. Table 3 Odds ratios (ORs) with their 95% Confidence Intervals (CIs) for AttentionDeficit/Hyperactivity Disorders (ADHD F90) in association with NOx exposure during fetal life in crude and adjusted models. Quartilesa

Crude modelb

Adjusted modelc

OR (95% CI)

OR (95% CI)

Trimester 1 Quartile 1a Quartile 2a Quartile 3a Quartile 4a Linear per 10 μg/m3

1 0.77 0.81 0.96 1.00

(0.62–0.95) (0.65–0.99) (0.79–1.17) (0.92–1.09)

1 0.99 0.95 1.01 1.01

(0.78–1.26) (0.74–1.21) (0.80–1.28) (0.91–1.11)

Trimester 2 Quartile 1a Quartile 2a Quartile 3a Quartile 4a Linear per 10 μg/m3

1 0.86 0.89 0.91 0.99

(0.70–1.06) (0.73–1.10) (0.74–1.12) (0.91–1.08)

1 1.06 1.04 0.96 1.01

(0.84–1.34) (0.81–1.32) (0.75–1.22) (0.92–1.12)

Trimester 3 Quartile 1a Quartile 2a Quartile 3a Quartile 4a Linear per 10 μg/m3

1 0.82 0.85 0.92 1.00

(0.66–1.01) (0.69–1.04) (0.75–1.12) (0.92–1.09)

1 1.02 (0.81–1.29) 0.92 (0.72–1.17) 1.00 (0.78–1.26) 1.02(0.92–1.13)

All pregnancyc Quartile 1a Quartile 2a Quartile 3a Quartile 4a Linear per 10 μg/m3

1 0.82 0.92 0.97 0.99

(0.67–1.01) (0.75–1.13) (0.80–1.19) (0.91–1.09)

1 1.04 1.07 1.06 1.02

4.2. Neuro-inflammation and important exposure periods A suggested pathway of air pollution's effect on the brain is the same one that effects the lungs and heart: inflammation and oxidative stress (Block and Calderon-Garciduenas, 2009). Children with autism present more neuro-inflammation and systemic inflammation, both previously linked to air pollution exposure (Costa et al., 2017). Accumulating evidence from animal studies indicate that air pollution exposure during the gestational period is linked to developmental neurotoxicity in mice (Allen et al., 2017). Indeed, the association between prenatal air pollution and ASD is supported by a recent animal study showing ASD traits in mice after gestational diesel exposure (Chang et al., 2018). The most important exposure period in animal models seems to correspond to third trimester exposure in humans (Allen et al., 2017). This is in line with previous epidemiological studies that illustrate more evident air pollution effects in humans during the third trimester (Kalkbrenner et al., 2015; Raz et al., 2015; Costa et al., 2017). Our estimated associations during the prenatal period also indicate a stronger effect in the third trimester. The high correlation of exposure between trimester-specific NOx values should, however, be considered when interpreting trimester-specific results. Other studies have instead found the postnatal period to be more relevant when adjusting for both prenatal and postnatal time-periods (Volk et al., 2013; Raz et al., 2018; Ritz et al., 2018). Unfortunately, we could not study this aspect in our cohort because we did not have information on residential addresses during childhood.

(0.82–1.32) (0.85–1.36) (0.83–1.35) (0.92–1.13)

4.3. In relation to previous literature on ASD and air pollution

Year-specific (birth year) quartiles with first quartile as reference. The linear estimates are adjusted for birth year (since the quartiles are birth year-specific). c Variables included into the adjusted models: gender of child, maternal age, parity, maternal smoking, maternal BMI, maternal education, disposable income, maternal country of birth. a b

Our study is in agreement with most studies regarding the positive association of air pollution and ASD found (Jung et al., 2013; Talbott et al., 2015; Goodrich et al., 2018; Kerin et al., 2018; Raz et al., 2018). The European studies not finding associations could possibly have been subjected to outcome misclassification, as reported autistic traits were used instead of clinical diagnoses (Gong et al., 2014; Guxens et al., 2016). In fact, air pollution was found to have an association on clinically diagnosed autism in Denmark (Ritz et al., 2018). A previous Swedish study based in Stockholm, however, did not find an effect when using clinical diagnoses (Gong et al., 2017). As previous studies have found links between ASD and SES (Rai et al., 2012) and maternal country of origin (Haglund and Kallen, 2011), careful adjustments of these factors have been carried out. However, the confounding effect of SES was small in the previous Swedish study (Gong et al., 2017) as well as in this present study. Another caveat mentioned in the literature is

4. Discussion 4.1. Complexity of ASD etiology The complex etiology of ASD has been discussed over the last decades, especially with the hope of understanding the substantial increase of both children and adults diagnosed with autism. To date, heredity and genetics constitute the major risk factors for ASD, with having an older sibling diagnosed with ASD being the most common determinant

5

Environment International 133 (2019) 105149

A. Oudin, et al.

confounded by behavior (as opposed to using personal air pollution measurements). We observed a linear dose-response curve until levels up to around 25 μg/m3, where the association seemed to level off or decrease. We cannot know for sure, but this could be due to unmeasured residual confounding in Malmö, the main city in the study area. The statistical power is unfortunately too low to study associations above 25 μg/m3 separately. In our cohort study, we used logistic regression analysis to estimate the associations between prenatal exposure to air pollution and risk of ADHD and ASD. Although time-to-event analysis, specifically Cox regression, may be a somewhat standard method for an open cohort study design, the relevance of using a time-to-event analysis method could be questioned when it comes to congenital disorders. Time to diagnosis may depend on other factors than those studied, for example socioeconomic factors, which may in turn be associated with the exposure of interest (here: air pollution). If this aspect of using time-to-event analysis would introduce bias in our setting, we doubt that it would explain the associations. On the contrary, it may hide them since children living in socially less privileged areas will be diagnosed later given the same severity of illness. Furthermore, in a cohort study, the odds ratios produced by logistic regression should approximate the Hazard Ratios from a Cox Proportional Hazard Regression. As a sensitivity analysis, we therefore ran all models with Cox Proportional Hazard Regression, and the results were indeed very similar.

the impact other co-morbidities have on air pollution and ASD risk. It has been suggested that a child with ADHD or intellectual developmental disorders are more likely to be diagnosed with ASD than a child without such co-morbidities (Gillberg and Fernell, 2014). In turn, clinical diagnosis of ASD might entail greater support in society and within schools, while other co-morbidities alone might not involve the same level of assistance, hence there might be some misclassification (Gillberg and Fernell, 2014). We could partially adjust for this possible misclassification by excluding children also diagnosed with ADHD when studying ASD, but we lacked data on intellectual development disorders. However, we could study cases of Childhood autism with more strict criteria, and in doing so, even moderately stronger associations were identified. 4.4. In relation to previous literature on ADHD and air pollution In not providing any associations between diagnosed ADHD and air pollution, our study is in agreement with published articles on observed ADHD symptoms (Gong et al., 2014; Forns et al., 2018; Myhre et al., 2018). On the contrary, two Japanese studies did find a link between ADHD-like behavior and air pollution, but they utilized coarse spatial resolutions and lacked clinical diagnoses (Yorifuji et al., 2017). Studies in Spain and the US found prenatal exposure to air pollution to be associated with poorer scores in attention functions (Chiu et al., 2016; Sentis et al., 2017). Furthermore, a German study revealed an association between ADHD symptoms and air pollution levels during late childhood years (Fuertes et al., 2016). Exposure to higher polycyclic aromatic hydrocarbon (PAH) levels was even associated with ADHD symptoms in a US study (Perera et al., 2018). Here, children whose mothers suffered economic hardships were more susceptible to the effects of air pollution, illustrating the impact of SES (Perera et al., 2018). Our study used clinical diagnoses of ADHD instead of symptoms, which adheres to recommended methods (Myhre et al., 2018). We have only identified one other study using clinical diagnoses of ADHD and air pollution. The authors of this South Korean-based study found associations between ADHD and cumulative air pollution exposures from birth to time of diagnosis (Min and Min, 2017). The discrepancy between their results and ours could be attributed to the differences in exposure assessments or exposure levels.

4.6. Public health implication This study contributes to the growing evidence of a link between prenatal exposure to air pollution and ASD. Autism can have a significant influence on both the individual's and family's wellbeing as well as future economic stability and productivity. Thus, we as a society should consider all potential risk factors associated with the development of ASD and actively work to mitigate and prevent them. Additionally, air pollution has been shown to be unevenly distributed among different socioeconomic groups, even in a relatively egalitarian country such as Sweden (Stroh et al., 2005). It has been postulated that if air pollution impairs children's health, ability to learn, and potential to contribute to society, further segregation between communities will occur (Perera, 2017). The environment is of special importance to the unborn child because of its underdeveloped defenses and rapid neurodevelopment. Today, bold policies and new technology are lowering harmful emissions, decreasing average exposure levels, and contributing to a safer environment for all children. Indeed, scientific evidence has shown that interventions to lower emissions can benefit children's neurodevelopment (Kalia et al., 2017). Moreover, a recent study has given indication that folic acid intake during pregnancy could mitigate air pollution's negative effects (Goodrich et al., 2018). More studies on possible preventive measures on both an individual and societal level are urgently needed. The wellbeing of our children as a shared fundamental value, irrespective of cultures and borders, presents a politically powerful catalyst for action to lower pollutant levels (Perera, 2017).

4.5. Exposure and methodological considerations We used NOx, a well-known marker for near-roadway exposure to air pollutants or ultrafine particles, to assess exposure (Arhami et al., 2009). Ultrafine particles could reach the brain directly through the olfactory bulb or indirectly via inhalation into the lungs and transportation by blood circulation (Block and Calderon-Garciduenas, 2009; Lucchini et al., 2012). Naturally then, the available pathway for prenatal exposure to fine particles would be the circulatory system, with air pollutants inhaled by the mother ultimately traveling through the umbilical cord to the fetus. NOx on its own could also have an adverse effect on brain function, as it is a potent oxidant. For instance, animal studies involving rats have shown that nitrogen dioxide, the largest component of NOx, can cause mitochondrial damage and harm the brain (Yan et al., 2015). A previous study investigating air toxics identified traffic-related air toxics to be associated with ASD diagnoses and severity of disorder (Kalkbrenner et al., 2018). A weakness in our study regarding exposure classification could be the exclusion of any source-specific exposure; however, NOx exposure correlates well with traffic-related emissions (Cyrys et al., 2012). As in most other epidemiological air pollution studies, we assessed exposure at each participant's home residence. This means that we, due to lack of such data, disregarded other sources of exposure, such as exposure during commuting or at workand maternal behavior. An advantage with an exposure assessment model that does not take maternal behavior into account is that exposure assessment cannot be associated with or

5. Conclusions In this longitudinal cohort study, we found positive associations between air pollution exposure during the prenatal period and an increased risk of developing ASD. We did not find similar associations between air pollution and the risk of developing ADHD. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. 6

Environment International 133 (2019) 105149

A. Oudin, et al.

Acknowledgment

Verhulst, A. von Berg, T. Vrijkotte, A.-M. Nybo Andersen, B. Heude, U. Krämer, J. Heinrich, H. Tiemeier, F. Forastiere, G. Pershagen, B. Brunekreef and M. Guxens (2018). "Air pollution exposure during pregnancy and symptoms of attention deficit and hyperactivity disorder in children in Europe." Epidemiology 29(5): 618–626. Frondelius, K., Oudin, A., Malmqvist, E., 2018. Traffic-related air pollution and child BMI—a study of prenatal exposure to nitrogen oxides and body mass index in children at the age of four years in Malmö, Sweden. Int. J. Environ. Res. Public Health 15 (10), 2294. Fuertes, E., Standl, M., Forns, J., Berdel, D., Garcia-Aymerich, J., Markevych, I., SchulteKoerne, G., Sugiri, D., Schikowski, T., Tiesler, C.M., Heinrich, J., 2016. Traffic-related air pollution and hyperactivity/inattention, dyslexia and dyscalculia in adolescents of the German GINIplus and LISAplus birth cohorts. Environ. Int. 97, 85–92. Gillberg, C., Fernell, E., 2014. Autism plus versus autism pure. J. Autism Dev. Disord. 44 (12), 3274–3276. Gong, T., Almqvist, C., Bolte, S., Lichtenstein, P., Anckarsater, H., Lind, T., Lundholm, C., Pershagen, G., 2014. Exposure to air pollution from traffic and neurodevelopmental disorders in Swedish twins. Twin Res Hum Genet 17 (6), 553–562. Gong, T., Dalman, C., Wicks, S., Dal, H., Magnusson, C., Lundholm, C., Almqvist, C., Pershagen, G., 2017. Perinatal exposure to traffic-related air pollution and autism spectrum disorders. Environ. Health Perspect. 125 (1), 119–126. Goodrich, A.J., Volk, H.E., Tancredi, D.J., McConnell, R., Lurmann, F.W., Hansen, R.L., Schmidt, R.J., 2018. Joint effects of prenatal air pollutant exposure and maternal folic acid supplementation on risk of autism spectrum disorder. Autism Res. 11 (1), 69–80. Guxens, M., Ghassabian, A., Gong, T., Garcia-Esteban, R., Porta, D., Giorgis-Allemand, L., Almqvist, C., Aranbarri, A., Beelen, R., Badaloni, C., Cesaroni, G., de Nazelle, A., Estarlich, M., Forastiere, F., Forns, J., Gehring, U., Ibarluzea, J., Jaddoe, V.W., Korek, M., Lichtenstein, P., Nieuwenhuijsen, M.J., Rebagliato, M., Slama, R., Tiemeier, H., Verhulst, F.C., Volk, H.E., Pershagen, G., Brunekreef, B., Sunyer, J., 2016. Air pollution exposure during pregnancy and childhood autistic traits in four European population-based cohort studies: the ESCAPE project. Environ. Health Perspect. 124 (1), 133–140. Haglund, N.G., Kallen, K.B., 2011. Risk factors for autism and Asperger syndrome. Perinatal factors and migration. Autism 15 (2), 163–183. Hertz-Picciotto, I., Croen, L.A., Hansen, R., Jones, C.R., van de Water, J., Pessah, I.N., 2006. The CHARGE study: an epidemiologic investigation of genetic and environmental factors contributing to autism. Environ. Health Perspect. 114 (7), 1119–1125. Ivert, A.K., Merlo, J., Svensson, R., Levander, M.T., 2013. How are immigrant background and gender associated with the utilisation of psychiatric care among adolescents? Soc. Psychiatry Psychiatr. Epidemiol. 48 (5), 693–699. Jung, C.R., Lin, Y.T., Hwang, B.F., 2013. Air pollution and newly diagnostic autism spectrum disorders: a population-based cohort study in Taiwan. PLoS One 8 (9), e75510. Kalia, V., Perera, F., Tang, D., 2017. Environmental pollutants and neurodevelopment: review of benefits from closure of a coal-burning power plant in Tongliang, China. Glob Pediatr Health 4 (2333794x17721609). Kalkbrenner, A.E., Windham, G.C., Serre, M.L., Akita, Y., Wang, X., Hoffman, K., Thayer, B.P., Daniels, J.L., 2015. Particulate matter exposure, prenatal and postnatal windows of susceptibility, and autism spectrum disorders. Epidemiology 26 (1), 30–42. Kalkbrenner, A.E., Windham, G.C., Zheng, C., McConnell, R., Lee, N.L., Schauer, J.J., Thayer, B., Pandey, J., Volk, H.E., 2018. Air toxics in relation to autism diagnosis, phenotype, and severity in a U.S. Family-based study. Environ. Health Perspect. 126 (3), 037004. Kerin, T., Volk, H., Li, W., Lurmann, F., Eckel, S., McConnell, R., Hertz-Picciotto, I., 2018. Association between air pollution exposure, cognitive and adaptive function, and ASD severity among children with autism spectrum disorder. J. Autism Dev. Disord. 48 (1), 137–150. Klassen, A.F., Miller, A., Fine, S., 2004. Health-related quality of life in children and adolescents who have a diagnosis of attention-deficit/hyperactivity disorder. Pediatrics 114 (5), e541–e547. Lai, M.-C., Lombardo, M.V., Baron-Cohen, S., 2014. Autism. Lancet 383 (9920), 896–910. Lee, L.C., Harrington, R.A., Louie, B.B., Newschaffer, C.J., 2008. Children with autism: quality of life and parental concerns. J. Autism Dev. Disord. 38 (6), 1147–1160. Lord, C., Rutter, M., Le Couteur, A., 1994. Autism Diagnostic Interview-Revised: a revised version of a diagnostic interview for caregivers of individuals with possible pervasive developmental disorders. J. Autism Dev. Disord. 24 (5), 659–685. Lord, C., Risi, S., Lambrecht, L., Cook Jr., E.H., Leventhal, B.L., DiLavore, P.C., Pickles, A., Rutter, M., 2000. The autism diagnostic observation schedule-generic: a standard measure of social and communication deficits associated with the spectrum of autism. J. Autism Dev. Disord. 30 (3), 205–223. Lucchini, R.G., Dorman, D.C., Elder, A., Veronesi, B., 2012. Neurological impacts from inhalation of pollutants and the nose-brain connection. Neurotoxicology 33 (4), 838–841. Malmqvist, E., Jakobsson, K., Tinnerberg, H., Rignell-Hydbom, A., Rylander, L., 2013. Gestational diabetes and preeclampsia in association with air pollution at levels below current air quality guidelines. Environ. Health Perspect. 121 (4), 488–493. Malmqvist, E., Larsson, H.E., Jonsson, I., Rignell-Hydbom, A., Ivarsson, S.A., Tinnerberg, H., Stroh, E., Rittner, R., Jakobsson, K., Swietlicki, E., Rylander, L., 2015. Maternal exposure to air pollution and type 1 diabetes–accounting for genetic factors. Environ. Res. 140, 268–274. Malmqvist, E., Liew, Z., Kallen, K., Rignell-Hydbom, A., Rittner, R., Rylander, L., Ritz, B., 2017. Fetal growth and air pollution - a study on ultrasound and birth measures. Environ. Res. 152, 73–80. Min, J.Y., Min, K.B., 2017. Exposure to ambient PM10 and NO2 and the incidence of attention-deficit hyperactivity disorder in childhood. Environ. Int. 99, 221–227. Muhle, R., Trentacoste, S.V., Rapin, I., 2004. The genetics of autism. Pediatrics 113 (5),

We are grateful to Swedish Research Council FORTE for funding this research (grant number 2015-00923). We are also grateful to Prof. Beate Ritz for valuable input and to Erin Flanagan for help editing. Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.envint.2019.105149. References Allen, J.L., Oberdorster, G., Morris-Schafer, K., Wong, C., Klocke, C., Sobolewski, M., Conrad, K., Mayer-Proschel, M., Cory-Slechta, D.A., 2017. Developmental neurotoxicity of inhaled ambient ultrafine particle air pollution: parallels with neuropathological and behavioral features of autism and other neurodevelopmental disorders. Neurotoxicology 59, 140–154. Arhami, M., Polidori, A., Delfino, R.J., Tjoa, T., Sioutas, C., 2009. Associations between personal, indoor, and residential outdoor pollutant concentrations: implications for exposure assessment to size-fractionated particulate matter. J Air Waste Manag Assoc 59 (4), 392–404. Arnold, L.E., 1996. Sex differences in ADHD: conference summary. J. Abnorm. Child Psychol. 24 (5), 555–569. Arvidsson, O., Gillberg, C., Lichtenstein, P., Lundstrom, S., 2018. Secular changes in the symptom level of clinically diagnosed autism. J. Child Psychol. Psychiatry 59 (7), 744–751. Block, M.L., Calderon-Garciduenas, L., 2009. Air pollution: mechanisms of neuroinflammation and CNS disease. Trends Neurosci. 32 (9), 506–516. Bolton, J.L., Smith, S.H., Huff, N.C., Gilmour, M.I., Foster, W.M., Auten, R.L., Bilbo, S.D., 2012. Prenatal air pollution exposure induces neuroinflammation and predisposes offspring to weight gain in adulthood in a sex-specific manner. FASEB J. 26 (11), 4743–4754. Calderon-Garciduenas, L., Kulesza, R.J., Doty, R.L., D’Angiulli, A., Torres-Jardon, R., 2015a. Megacities air pollution problems: Mexico City Metropolitan Area critical issues on the central nervous system pediatric impact. Environ. Res. 137, 157–169. Calderon-Garciduenas, L., Vojdani, A., Blaurock-Busch, E., Busch, Y., Friedle, A., FrancoLira, M., Sarathi-Mukherjee, P., Martinez-Aguirre, X., Park, S.B., Torres-Jardon, R., D’Angiulli, A., 2015b. Air pollution and children: neural and tight junction antibodies and combustion metals, the role of barrier breakdown and brain immunity in neurodegeneration. J. Alzheimers Dis. 43 (3), 1039–1058. Chang, Y.C., Cole, T.B., Costa, L.G., 2018. Prenatal and early-life diesel exhaust exposure causes autism-like behavioral changes in mice. Part Fibre Toxicol 15 (1), 18. Chiu, Y.-H.M., Hsu, H.-H.L., Coull, B.A., Bellinger, D.C., Kloog, I., Schwartz, J., Wright, R.O., Wright, R.J., 2016. Prenatal particulate air pollution and neurodevelopment in urban children: examining sensitive windows and sex-specific associations. Environ. Int. 87, 56–65. Cohen, A.J., Brauer, M., Burnett, R., Anderson, H.R., Frostad, J., Estep, K., Balakrishnan, K., Brunekreef, B., Dandona, L., Dandona, R., Feigin, V., Freedman, G., Hubbell, B., Jobling, A., Kan, H., Knibbs, L., Liu, Y., Martin, R., Morawska, L., Pope, C.A., Shin, H., Straif, K., Shaddick, G., Thomas, M., van Dingenen, R., van Donkelaar, A., Vos, T., Murray, C.J.L., Forouzanfar, M.H., 2017. Estimates and 25-year trends of the global burden of disease attributable to ambient air pollution: an analysis of data from the Global Burden of Diseases Study 2015. Lancet 389 (10082), 1907–1918. Costa, L.G., Chang, Y.C., Cole, T.B., 2017. Developmental neurotoxicity of traffic-related air pollution: focus on autism. Curr Environ Health Rep 4 (2), 156–165. Craig, F., Lamanna, A.L., Margari, F., Matera, E., Simone, M., Margari, L., 2015. Overlap between autism spectrum disorders and attention deficit hyperactivity disorder: searching for distinctive/common clinical features. Autism Res. 8 (3), 328–337. Cyrys, J., Eeftens, M., Heinrich, J., Ampe, C., Armengaud, A., Beelen, R., Bellander, T., Beregszaszi, T., Birk, M., Cesaroni, G., Cirach, M., de Hoogh, K., De Nazelle, A., de Vocht, F., Declercq, C., Dėdelė, A., Dimakopoulou, K., Eriksen, K., Galassi, C., Grąulevičienė, R., Grivas, G., Gruzieva, O., Gustafsson, A.H., Hoffmann, B., Iakovides, M., Ineichen, A., Krämer, U., Lanki, T., Lozano, P., Madsen, C., Meliefste, K., Modig, L., Mölter, A., Mosler, G., Nieuwenhuijsen, M., Nonnemacher, M., Oldenwening, M., Peters, A., Pontet, S., Probst-Hensch, N., Quass, U., Raaschou-Nielsen, O., Ranzi, A., Sugiri, D., Stephanou, E.G., Taimisto, P., Tsai, M.-Y., Vaskövi, É., Villani, S., Wang, M., Brunekreef, B., Hoek, G., 2012. Variation of NO2 and NOx concentrations between and within 36 European study areas: results from the ESCAPE study. Atmos. Environ. 62, 374–390. Dachew, B.A., Mamun, A., Maravilla, J.C., Alati, R., 2018. Pre-eclampsia and the risk of autism-spectrum disorder in offspring: meta-analysis. Br. J. Psychiatry 212 (3), 142–147. Flores-Pajot, M.C., Ofner, M., Do, M.T., Lavigne, E., Villeneuve, P.J., 2016. Childhood autism spectrum disorders and exposure to nitrogen dioxide, and particulate matter air pollution: a review and meta-analysis. Environ. Res. 151, 763–776. Forns, J., J. Sunyer, R. Garcia-Esteban, D. Porta, A. Ghassabian, L. Giorgis-Allemand, T. Gong, U. Gehring, M. Sørensen, M. Standl, D. Sugiri, C. Almqvist, A. Andiarena, C. Badaloní, R. Beelen, D. Berdel, G. Cesaroni, M.-A. Charles, K. T. Eriksen, M. Estarlich, M. F. Fernandez, A. Forhan, V. W. V. Jaddoe, M. Korek, P. Lichtenstein, A. Lertxundi, M.-J. Lopez-Espinosa, I. Markevych, A. de Nazelle, O. Raaschou-Nielsen, M. Nieuwenhuijsen, R. Pérez-Lobato, C. Philippat, R. Slama, C. M. T. Tiesler, F. C.

7

Environment International 133 (2019) 105149

A. Oudin, et al.

Sentis, A., Sunyer, J., Dalmau-Bueno, A., Andiarena, A., Ballester, F., Cirach, M., Estarlich, M., Fernandez-Somoano, A., Ibarluzea, J., Iniguez, C., Lertxundi, A., Tardon, A., Nieuwenhuijsen, M., Vrijheid, M., Guxens, M., 2017. Prenatal and postnatal exposure to NO2 and child attentional function at 4-5years of age. Environ. Int. 106, 170–177. Stroh, E., Oudin, A., Gustafsson, S., Pilesjo, P., Harrie, L., Stromberg, U., Jakobsson, K., 2005. Are associations between socio-economic characteristics and exposure to air pollution a question of study area size? An example from Scania, Sweden. Int. J. Health Geogr. 4, 30. Stroh, E., Harrie, L., Gustafsson, S., 2007. A study of spatial resolution in pollution exposure modelling. Int. J. Health Geogr. 6, 19. Stroh, E., Rittner, R., Oudin, A., Ardo, J., Jakobsson, K., Bjork, J., Tinnerberg, H., 2012. Measured and modeled personal and environmental NO2 exposure. Popul. Health Metrics 10 (1), 10. Talbott, E. O., L. P. Marshall, J. R. Rager, V. C. Arena, R. K. Sharma and S. L. Stacy (2015). "Air toxics and the risk of autism spectrum disorder: the results of a population based case-control study in southwestern Pennsylvania." Environ. Health 14: 80. Thapar, A., Cooper, M., Eyre, O., Langley, K., 2013. What have we learnt about the causes of ADHD? J. Child Psychol. Psychiatry 54 (1), 3–16. van Heijst, B.F., Geurts, H.M., 2015. Quality of life in autism across the lifespan: a metaanalysis. Autism 19 (2), 158–167. Volk, H.E., Lurmann, F., Penfold, B., Hertz-Picciotto, I., McConnell, R., 2013. Trafficrelated air pollution, particulate matter, and autism. JAMA Psychiatry 70 (1), 71–77. Wang, C., Geng, H., Liu, W., Zhang, G., 2017. Prenatal, perinatal, and postnatal factors associated with autism: a meta-analysis. Medicine (Baltimore) 96 (18), e6696. Weisskopf, M.G., Kioumourtzoglou, M.A., Roberts, A.L., 2015. Air pollution and autism Spectrum disorders: causal or confounded? Curr Environ Health Rep 2 (4), 430–439. Werling, D.M., Geschwind, D.H., 2013. Sex differences in autism spectrum disorders. Curr. Opin. Neurol. 26 (2), 146–153. Yan, W., Ji, X., Shi, J., Li, G., Sang, N., 2015. Acute nitrogen dioxide inhalation induces mitochondrial dysfunction in rat brain. Environ. Res. 138, 416–424. Yang, C., Zhao, W., Deng, K., Zhou, V., Zhou, X., Hou, Y., 2017. The association between air pollutants and autism spectrum disorders. Environ. Sci. Pollut. Res. 24 (19), 15949–15958. Yorifuji, T., Kashima, S., Diez, M.H., Kado, Y., Sanada, S., Doi, H., 2017. Prenatal exposure to outdoor air pollution and child behavioral problems at school age in Japan. Environ. Int. 99, 192–198. Zhu, T., Gan, J., Huang, J., Li, Y., Qu, Y., Mu, D., 2016. Association between perinatal hypoxic-ischemic conditions and attention-deficit/hyperactivity disorder: a metaanalysis. J. Child Neurol. 31 (10), 1235–1244.

e472–e486. Myhre, O., Lag, M., Villanger, G.D., Oftedal, B., Ovrevik, J., Holme, J.A., Aase, H., Paulsen, R.E., Bal-Price, A., Dirven, H., 2018. Early life exposure to air pollution particulate matter (PM) as risk factor for attention deficit/hyperactivity disorder (ADHD): need for novel strategies for mechanisms and causalities. Toxicol. Appl. Pharmacol. 354, 196–214. Oudin, A., Stroh, E., Stromberg, U., Jakobsson, K., Bjork, J., 2009. Long-term exposure to air pollution and hospital admissions for ischemic stroke. A register-based casecontrol study using modelled NO(x) as exposure proxy. BMC Public Health 9, 301. Perera, F., 2017. Pollution from fossil-fuel combustion is the leading environmental threat to global pediatric health and equity: solutions exist. Int. J. Environ. Res. Public Health 15 (1). Perera, F.P., Tang, D., Wang, S., Vishnevetsky, J., Zhang, B., Diaz, D., Camann, D., Rauh, V., 2012. Prenatal polycyclic aromatic hydrocarbon (PAH) exposure and child behavior at age 6-7 years. Environ. Health Perspect. 120 (6), 921–926. Perera, F.P., Wheelock, K., Wang, Y., Tang, D., Margolis, A.E., Badia, G., Cowell, W., Miller, R.L., Rauh, V., Wang, S., Herbstman, J.B., 2018. Combined effects of prenatal exposure to polycyclic aromatic hydrocarbons and material hardship on child ADHD behavior problems. Environ. Res. 160, 506–513. Posar, A., Visconti, P., 2017. Autism in 2016: the need for answers. J. Pediatr. 93 (2), 111–119. Rai, D., Lewis, G., Lundberg, M., Araya, R., Svensson, A., Dalman, C., Carpenter, P., Magnusson, C., 2012. Parental socioeconomic status and risk of offspring autism spectrum disorders in a Swedish population-based study. J. Am. Acad. Child Adolesc. Psychiatry 51 (5), 467–476 (e466). Raz, R., Roberts, A.L., Lyall, K., Hart, J.E., Just, A.C., Laden, F., Weisskopf, M.G., 2015. Autism spectrum disorder and particulate matter air pollution before, during, and after pregnancy: a nested case-control analysis within the Nurses’ Health Study II Cohort. Environ. Health Perspect. 123 (3), 264–270. Raz, R., H. Levine, O. Pinto, D. M. Broday, Yuval and M. G. Weisskopf (2018). "Trafficrelated air pollution and autism spectrum disorder: a population-based nested casecontrol study in Israel." Am. J. Epidemiol. 187(4): 717–725. Ritz, B., Liew, Z., Yan, Q., Cui, X., Virk, J., Ketzel, M., Raaschou-Nielsen, O., 2018. Air pollution and autism in Denmark. Environ Epidemiol 2 (4). Rosen, B.N., Lee, B.K., Lee, N.L., Yang, Y., Burstyn, I., 2015. Maternal smoking and autism spectrum disorder: a meta-analysis. J. Autism Dev. Disord. 45 (6), 1689–1698. Sandin, S., Lichtenstein, P., Kuja-Halkola, R., Hultman, C., Larsson, H., Reichenberg, A., 2017. The heritability of autism Spectrum disorder. Jama 318 (12), 1182–1184. Sciberras, E., Mulraney, M., Silva, D., Coghill, D., 2017. Prenatal risk factors and the etiology of ADHD-review of existing evidence. Curr Psychiatry Rep 19 (1), 1.

8